The development and deployment of digital information networks is accompanied by new concerns for the protection of rights to data and information. The U.S. Congress Office of Technology Assessment identified the following key developments relevant to the area of this invention: there has been an overall movement to distributed computing; boundaries between types of information are blurring; the number and variety of service providers has increased. Information Security and Privacy in Networked Environments, Congress, Office of Technology Assessment, OTA-TCT-606, Washington, DC: U.S. Government Printing Office, September 1994.
Computer networks allow more interactivity; and, most significantly, electronic information has opened new questions about copyright, ownership, and responsibility for information. Technology, business practice, and law are changing at different rates, law arguably being the slowest.
Intellectual property, or information, is different from real property. A major difference between intellectual property and real property is that intellectual property can be embodied in forms which can be copied from the owner while the owner still retains the original. For example, a broadcast or performance of a musical composition can be recorded (and copies made of the recording) while the composer retains the original composition; a photograph can be reproduced while the owner retains the original negative.
In the past, when information was stored in analog form, the copying and redistribution of such information, while problematic, did not account for as much economic loss as is possible today. The storage of information in analog form uses a physical medium that is made to have some characteristic vary in proportion with the information to be stored. For instance, the groove on a vinyl record captures the frequency and intensity (volume) of a sound by the extent of its excursion. At each stage in the process of playing a record: the stylus tracing the groove, generation of a small voltage, amplification of the voltage, and reproduction of the sound, small errors are introduced. Today's high fidelity systems are very accurate, but they are not flawless.
Indeed, copying a vinyl record to a cassette tape results in a small, but noticeable, reduction in sound quality. If multiple generations of recording (e.g., cascaded recordings) were undertaken, the resulting product would be noticeably inferior to the original. Similarly, when multiple generations of photocopies of an image are made, the quality of the resulting image is typically poor, with many dark and light areas that were not present in the original image.
It is the inevitable gradual degradation of quality that has proven to be a practical disincentive to large scale copying of analog information. Notwithstanding this observation, where the potential profits are high, such copying is undertaken even though the resulting product's quality is significantly below that of the original. Videotape copies of movies represent a good example. Some fraction of the marketplace is willing to accept a lower quality product in exchange for a significantly lower price. The logistics associated with making large numbers of copies (an inherently serial process), including obtaining the raw materials (cassettes), the reproduction equipment, and the distribution channels also have served to limit illicit production. Finally, the quality of the product as well as the markings on the package distinguish it from the original and may also serve as a disincentive (for some) to purchase an illicit copy.
Just as the invention of the printing press changed the way in which society interacted with information on paper, the technical advances in digital computers and communications in the closing years of the twentieth century have a potential for high impact on legal, moral, and business practice. The printing press is often credited as an enabling mechanism for the Renaissance and the Reformation in Europe. The advances in digital information technology will similarly impact commerce and law. Digital technology enables changing the representation of information without changing the content. (Of course the content can be changed too.)
The storage of information in digital form depends on the ability to encode information in binary form to arbitrary precision and to record that binary form in a physical medium that can take on two distinct characteristics. Preserving the fidelity of information recorded in binary (using media with two distinct and easily-differentiated characteristics) is easily accomplished. For instance, a compact disc stores information (each binary digit or bit) as the presence or absence of a hole (depression or pit) that reflects or does not reflect light. Compared to the analog recording of phonograph records, the information stored in each hole is unambiguously a binary digit, the value of which is either zero or one. No other values are possible. A digital tape stores each bit as a magnetic spot that is oriented either north/south or south/north. Today's digital sound systems use sufficiently many bits to capture sound levels beyond the ability of the human ear to distinguish a difference and in so doing attain so-called "perfect" fidelity.
A digital file can be copied with no loss of fidelity (as the mechanism need only distinguish between two easily-differentiated states). With straightforward and well-known error-correction mechanisms, even inevitable flaws can be made so improbable as to occur fewer than once in ten billion bits.
As a result of the ability to copy a file with no loss of fidelity, it is now almost impossible to differentiate a digital copy from the digital original. In a network environment recording materials, reproduction equipment and distribution are not impediments to copying. Consequently, in the digital domain the threshold inhibiting the making of illicit copies is significantly lowered. Evidence that this is the case is presented by the Software Publishers Association and by the Business Software Alliance, each of which indicates that billions of dollars of software is pirated (in the sense of being illicitly copied) each year. Additionally, print publishers hesitate to expand into the network marketplace because they are unable to control (in the sense of receiving compensation in return for rights) secondary distribution of their products as well as incorporation of their products into derivative products. Digitally stored information may include binary data, computer software, text, graphics, audio, and video. The uses of this information include news, entertainment, education, and analysis. Information may be distributed in many ways, including networks, magnetic media, CD-ROM, semiconductor memory modules, and wireless broadcast.
Copying and distributing large volumes of digital information over long distances is becoming easier and less costly. Such changes in cost and convenience of necessity impact business decisions concerning producing, distributing, promoting, and marketing. The commercial relationship among information producers (such as authors, performers, and artists), distributors (such as publishers, promoters, and broadcasters), and consumers must change in response to the technology.
The law concerning intellectual property is in ferment. Major revisions in the laws regarding the protection of computer programs have been suggested. A Manifesto Concerning the Legal Protection of Computer Programs, Samuelson, P. R. et al., Columbia Law Review, vol. 94, no. 8, pp. 2308-2431, December 1994. The European Union is working on harmonizing protection of intellectual property rights with respect to technology and differences in civil and common law countries. Commission of the European Union, Jul. 19, 1995, Green Paper on Copyright and Neighboring Rights in the Information Society, catalogue number CB-CO-95-421-EN-C, ISSN 0254-1475, ISBM 92-77-92580-9, Office for Official Publications of the European Communities, L-2985 Luxembourg. In the United States, the issue of protection of intellectual property rights is being addressed in the context of the National Information Infrastructure. The uncertainty of legal protection over time and from country to country only serves to emphasize the importance of and need for technical protection of intellectual property rights in information and data.
The principal technology which has been used for protecting intellectual property is cryptography. However, devising practical retail systems for delivery of intellectual property from distributor to consumer, as distinct from confidential transmission in national security and business activities among trusted and cleared personnel, has required innovation.
Executable software-based cryptography can ensure that data are distributed only to authorized users. The information to be protected is encrypted and transmitted to the authorized user(s). Separately, a decryption key is provided only to authorized users. The key is subsequently used to enable decryption of the information so that it is available to the authorized user(s).
Other ways of controlling access to portions of data or software have included the use of external devices or tokens (dongles) needed in order to access the data or selected features of a program. Possession of the token is made evident to the computer system by physical attachment of the token to the computer. A token is generally attached to a printer, game, or network port where executable software can check on its presence prior to authorizing access. Diskettes have also been used as dongles; their presence in the diskette drive is checked by the executing software. Because they must be actively interrogated, dongles are generally used to limit access to program features and not to limit access to information.
Of those prior art systems which make some use of encryption, none protects the data after it has been decrypted. Thus, secondary distribution and multiple uses are possible.
Further, in all of the prior art, access is all or nothing, that is, once access is granted, it cannot be controlled in any other ways. This makes it difficult to control copying, secondary distribution, as well as to obtain payment for all uses.
Originator controlled data dissemination is desirable. Several policies for control of dissemination of paper documents are specified in Control of Dissemination of Intelligence Information, Directive No. 1/7, Director of Central Intelligence, 4 May 1981. This Originator-Controlled (ORCON) policy has motivated development of computerized access controls. ORCON requires the permission of the originator to distribute information beyond the original receivers designated by the originator. The Propagated Access Control (PAC) policy and the related Propagated Access Control List (PACL) were proposed as one way of implementing ORCON. "On the Need for a Third Form of Access Control," Graubart, R., Proceedings of the 12th National Computer Security Conference, pp. 296-303, 1989. Whenever an authorized subject reads an object with an associated PACL, that PACL becomes associated with the subject. Any new object created by the subject inherits the PACL. PACLs are associated with both subjects and objects.
Owner-Retained Access Control (ORAC) (described in "Beyond the Pale of MAC and DAC: Defining New Forms of Access Control," McCollum, C. J., et al. Proceedings of the Symposium on Research in Security and Privacy, IEEE Computer Society Press, 1990) is similar to PAC in propagating ACLs with non-discretionary enforcement. ORAC goes further, retaining the autonomy of all originators associated with a given object in making access decisions, while basing mediation of requests on the intersection of the access rights that have been granted. ORAC is motivated to implement several of the DCID 1/7 policies in addition to ORCON, namely NO_CONTRACTOR, NO_FOREIGN, and RELEASABLE_TO.
Originator-Controlled Access Control (ORGCON) (described in "Generalized Framework for Access Control: Towards Prototyping the ORGCON Policy," Abrams, M. D., et al. Proceedings of the 14th National Computer Security Conference, October 1991) is a strong form of identity-based access control--it explicitly defines authority and delegation of authority, provides for accountability, and has an explicit inheritance policy. In ORGCON, the distribution list is indelibly attached to the object (i.e., the distribution list cannot be disassociated from the object, even in the limited cases where copying is permitted). ORGCON is a read, no-copy policy. Its formal model (taught in "A Rule-Set Approach to Formal Modeling of a Trusted Computer System," LaPadula, L. J., Computing Systems Journal, Vol. 7, No. 1, pp. 113-167, Winter 1994) distinguishes among device types in order to deal with the policy that no storage copy of an object is permitted. Information may be copied only to the display and printer, but not to any other device types.
The Typed Access Matrix (TAM) Model (described in "The Typed Access Matrix Model," Sandhu, R. S., Proceedings of the Symposium on Research in Security and Privacy, IEEE Computer society, pp. 122-136, 1992; and "Implementation Considerations for the Typed Access Matrix Model in a Distributed Environment," Sandhu, R. S., and G. S. Suri, 1992, Proceedings of the 15th National Computer Security Conference, pp. 221-235) incorporates strong typing into the access matrix model to provide a flexible model that can express a rich variety of security policies while addressing propagation of access rights and the safety problem. The safety problem is closely related to the fundamental flaw in Discretionary Access Control (DAC) that malicious code can modify the protection state. Types and rights are specified as part of the system definition; they are not predetermined in TAM.
The prior art, including cryptographic processes, tokens, dongles, so-balled "uncopyable" media, various executable software protection schemes, and executable software for printing that places an identifier on all printed output in a fashion not apparent to a human, fails to limit either secondary distribution or distribution of derivative works.
This shortcoming is not a failure of mechanism, but rather it is an architectural design omission. The problem of copying by the authorized user is simply not addressed. In each case, once the data are available to an authorized user, they are basically unprotected and may be copied, modified, or transmitted at will. Schemes that include identifiers on printed material, although they may aid in identifying the source of copied material, do not prevent secondary distribution.
Executable software-based cryptography can ensure that data are distributed only to authorized users. However, once data are received they may be freely manipulated and redistributed.
The information to be protected is encrypted and transmitted to the authorized user(s). In some systems the encrypted information is made freely available. Separately, a decryption key is provided only to authorized users. The key is subsequently used to enable decryption of the information so that it is available to the authorized user(s). It is at this point that the information is subject to manipulation and redistribution without further limitation.
As mentioned above, a dongle or token can be used to authorize access to executable software. However, once access has been granted to information that information is subject to manipulation and redistribution without further limitation. Further, dongles have proven to be unpopular because of the need to keep track of them and ensure that they are separately secured.
Uncopyable media, generally used either to control distribution of information or to control usage of executable software, are unpopular because of the user's inability to create a backup copy. Further, most so-called uncopyable disks have fallen victim to general-purpose duplication programs, rendering their protection useless. Sometimes, as in early releases of Lotus 1-2-3, an uncopyable disk was provided with the executable software release and had to be inserted in a floppy-disk drive for the executable software to function (operating as a disk dongle). Users soon learned how to by-pass the executable software so that the disk need not be present. Even where partially effective, the uncopyable disk did not serve as a deterrent to capturing information and redistributing it.
The degree of protection of data is typically made by the data owners and/or distributors based on their security analysis. It is common to perform security analysis in terms of risks, threats, vulnerabilities, and countermeasures. An owner's estimate of the probability that a particular threat will materialize is crucial to selecting appropriate rules to protect property rights.
Threat can be characterized as the intensity of attack on the data, which can be described as low, medium, and high.
Low For a security function to be rated as "suitable for use in a low threat environment," it shall be shown that the security function provides protection against unintended or casual breach of security by attackers possessing a low level of expertise, opportunities, resources and motivation. However, such a security function may be capable of being defeated by a knowledgeable attacker. Medium For a security function to be rated as "suitable for use in a medium threat environment," it shall be shown that the security function provides protection against attackers possessing a moderate level of expertise, opportunities, resources and motivation. High For a security function to be rated as "suitable for use in a high threat environment," it shall be shown that the security function provides protection against attackers possessing a high level of expertise, opportunity, resources and motivation. A successful attack is judged as being beyond normal practicality.
The following list covers some common anticipated threats to data and processing systems.
Threat: Capture of Output Signal
No matter what method is used to protect a data file, the data stored therein can be captured as a signal en route to an output device. Capture of an analog output results in some degradation of signal quality. But the market for bootleg copies of videos, for example, appears to be insensitive to such quality if the price is right. A captured digital signal suffers degradation of quality only as a result of bit errors (i.e., if the data capture was not completely accurate).
This threat is well known to the entertainment industry. Various approaches to protection have been incorporated in set-top boxes discussed in "Inside the Set-Top Box," Ciciora, W. S., IEEE Spectrum, pp. 70-75, April 1995.
Threat: Digital Copying
Once data have been decrypted, the resulting cleartext must be protected from unauthorized copying. Creating an unauthorized local copy, or disseminating the data without authorization each results in an original-quality copy without compensation to the owner.
Threat: Deliberate Attack Via Legacy (Pre-Existing) and Customized Hardware
High-intensity attack by attackers possessing a high level of expertise, opportunity, resources and motivation must be considered. Attackers in this category might include foreign governments and industrial espionage agents, teenage crackers, and resellers of pirated intellectual property. One manifestation of this threat is in uncontrolled hardware. The nominally protected information would be available in the memory and could be accessed via dual-ported memory or even by DMA (direct memory access) from a peripheral.
A strong indication of the usefulness and desirability of the present invention can be found in the legislation pending before the U.S. Congress to make illegal the by-passing or avoiding of copyright protection schemes. See S.1284, 104th Congress, 1st sess. (1995).
It is desirable to have a system of distributing data (intellectual property) that prevents copying, restricts re-distribution of the data and provides controlled access to the data.